association rules optimization: a survey - semantic · pdf fileassociation rules optimization:...

International Journal of Advanced Computer Research (ISSN (print): 2249-7277 ISSN (online): 2277-7970)

Volume-3 Number-1 Issue-9 March-2013

111

Association Rules Optimization: A Survey

Anshuman Singh Sadh1, Nitin Shukla

2

Abstract

Association Rule mining is one of the important and

most popular data mining technique. Association

rule mining can be efficiently used in any decision

making processor decision based rule generation. In

data mining task in general we will find the

frequent patterns to know the effective patterns

from the huge data. Then we find positive and

negative rules. If we observe the above phenomena

then we come to the point that the rule generation is

also huge. In this paper we survey several aspects of

optimization techniques by which we can optimize

the association rules. So the main motivation of our

survey is to minimize the rule generation or

optimize rule generation larger size of rules can be

minimized.

Keywords

Association Rule Mining, Positive Association, Negative

Association, Optimization

1. Introduction

Data Mining is expected to relieve current mining

methods from the sequential bottleneck, and provide

the ability to scale to massive data sets and improve

the response time [1]. Data mining has been a

powerful technique in analyzing and utilizing data in

today’s information-rich society. However, privacy is

nowadays a major concern in data mining

applications, which has led to a new research area,

privacy preserving data mining. A large amount of

research work has been devoted to this area, and

resulted in such techniques as k-anonymity [2], data

perturbation [3], [4], [5], [6], and data mining based

on [7], [8].

Association Rule Mining (ARM) is one of the most

used research are in data mining. ARM can be used

for discovering hidden relationship between items .

By given a user-specified threshold, also known as

minimum support, the mining of association rules can

discover the complete set of frequent patterns.

Anshuman Singh Sadh, M.Tech Scholar, SRIT , Jabalpur

(M.P), [email protected]

Nitin Shukla, Assistant Professor, SRIT, Jabalpur (M.P), [email protected]

That is, once the minimum support is given, the

complete set of frequent patterns is determined [9]. In

order to retrieve more correlations among items,

users may specify a relatively lower minimum

support[9]. Such a lower support often generates a

huge amount of frequent patterns; but most of the

patterns are already known or not interested to users.

It is a tedious task for users to filter out these

valueless patterns.

ARM is also studied in terms of market basket

analysis, which is the analysis of the itemset which

can be analyzed after the customer purchasing in the

mall [9]. It is just like the analysis of the customer of

purchasing behavior. Association rules also used in

various areas such as telecommunication networks,

market, risk management and inventory control etc.

[9][10].

In [11] author suggests that Data mining [1] is used

everywhere and large amounts of information are

gathered: in business, to analyses client behavior or

optimize production and sales [2].This signifies the

research direction in several fields. We can use ARM

and data mining application in health care, medical

database, classification and combining these

techniques with other approach extensively increases

the potential behavior and applicability.

Our papers main motivation to survey in the direction

of rule generation that is positive and negative

association rule. We also emphasize to optimize the

association rule so that it saves the time and un-

wanted rules can be avoided.

In [12] author suggests that many of the researchers

are generally focused on finding the positive rules

only but they not find the negative association rules.

But it is also important in analysis of intelligent data.

It works in the opposite manner of positive rule

finding. But problem with the negative association

rule is it uses large space and can take more time to

generate the rules as compare to the traditional

mining association rule [12]. So better optimization

technique can find a better solution in the above

direction.

We provide here a brief survey on ARM. Other

sections are arranged in the following manner:

Section 2 introduces negative and positive



112

association rules; Section 3 discusses about

optimization techniques; Section 4 describes about

Literature Review; section 5 shows the analysis;

Section 6 describes Conclusion.

2. Rule Generation

Association rule mining can be represent in terms of

A ⇒ B [S, C] where A and B are sets of items; S is

the support of the rules, defined as the rate of the

transactions containing all items in A and all items in

B i.e. Support (A ⇒ B) = P (A ∪ B) and C is the

confidence of the rule, defined as the ratio of S with

the rate of transactions containing A i.e. P (B / A).

Support and confidence are measures of the

interestingness of the rule. We calculate the support

value for justifying the usefulness of the items

present in the data set. A Higher support value

indicates the effectiveness for the enterprise. The

confidence signifies the decision theory, higher the

confidence higher the decision accuracy.

In the application domain, items correspond to web

resources, while transactions correspond to user

sessions. So the equality of rule signifies that the

domain is same and it can acquire the same alike set

in the frequent pattern generation. The basic

algorithm called Apriori Algorithm for finding the

association rules was proposed and later modified

uses Breadth First Search, Bottom Up Approach and

performs well when the Frequent Items are short and

and the use is easy. In this term we also discuss the

term candidate generation like in the apriori

algorithm, where it can generate 1-itemset,2-itemset

and n itemset according to the data set present.

As an example of an association rule, we can think of

the case of a super market. An association rule might,

then, be in the following form: 'If a customer buys

pork steaks, he buys at least two bottles of Coke as

well'. In other words, it is in the form X => Y, where

X and Y are items or set of items from the super

market’s database. In the case of City University, an

example of an association rule might be that if a

student wishes to do business studies, then with a

probability of 90% chooses City.

The problem of identifying association rules was first

introduced in (Agrawal, 1993). In (Hipp and Guntzer

and Nakhaeizadeh, 2000) the formal description of

the problem is given as follows: "Let ;={x1,...,xn} be

a set of distinct literals, called items. A set X⊆X;

with k=|X| is called a k-itemset or simply an itemset.

Let a database D be a multi-set of subsets. Each T∈ D

is called a transaction. We say that a transaction T∈

D supports an itemset X⊆X ; if X ⊆T holds. An

association rule is an expression X =>Y, where X,Y

are itemsets and X ∩Y = Ø holds. The fraction of

transactions T supporting an itemset X with respect

to database ' is called the support of X, supp(X)=

|{T∈D | X ⊆T}| / |D|. The support of a rule X =>Y is

defined as supp(X =>Y) = supp(X UY). The

confidence of this rule is defined as conf(X =>Y) =

supp(X U Y)/ supp(X)". The latter implies that we

are looking at the fraction of transactions that contain

the X itemset to see how many contain the Y itemset

as well.

3. Optimization Techniques

There are several optimization techniques which can

be apply on the association rule mining. Some of

techniques are as follows:

Ant Colony Optimization

The Ant Colony Optimization algorithm is mainly

inspired by the experiments run by Goss et al. [13]

which using a grouping of real ants in the real

environment. They study and observe the behavior of

those real ants and suggest that the real ants were

able to select the shortest path between their nest and

food resource, in the existence of alternate paths

between the two. This ant behavior was first

formulated and arranged as Ant System (AS) by

Dorigo et al. [14][15]. Based on the AS algorithm,

the Ant Colony Optimization (ACO) algorithm was

proposed [16]. In ACO algorithm, the optimization

problem can be expressed as a formulated graph G =

(C; L), where C is the set of components of the

problem, and L is the set of possible connections or

transitions among the elements of C.

Particle Swarm Optimization

PSO is a global optimization search algorithm

introduced by Kennedy and Eberhart. It is an

evolutionary computation technique discovered

through simulation of collective social behavior such

as bird flocking and fish schooling [17]. In PSO,

particles represent candidate solutions in a solution

space, and the optimal solution is found through

moving the particles in the solution space. Individual

particle flies through S-dimensional search space

with velocity dynamically adjusted according to its

own flying experience and its group’s flying

experience. The velocity and position of the particles

are adjusted according to the following equation:

Vid(t)=W.Vid(t-1+C1.R1.(Pid(t-1)-Xid(t-1)) +……+….

Xid(t)=Xid(t-1) + Vid(t-1)



113

Where d represents the value 1,2,3…. t represents the

looping; Vi and Xi are the velocity and position of

the ith particle; Pi is the previous personal best

position of particle i and is called pbest; Pg is the

previous best position of all the particles and is called

gbest; W is the inertia weight; C1 and C2 are positive

constants, and R1, R2.

Genetic Algorithm

Genetic algorithms work with a population of the

potential solutions[18]. In computing terms, genetic

algorithms map strings of numbers to each potential

solution. Each solution becomes an individual in the

population, and each string becomes a representation

of an individual [18]. There should be a way to derive

each individual from its string representation. The

genetic algorithm then manipulates the most

promising strings in its search for an improved

solution. This algorithm follows the following cycle.

1. Creation of a population of strings.

2. Evaluation of each string.

3. Selection of the best strings.

4. Genetic manipulation to create a new population of

strings.

4. Literature Review

In 2010 Ashutosh Dubey et al. [19] proposed a novel

data mining algorithm named J2ME-based Mobile

Progressive Pattern Mine (J2MPP-Mine) for effective

mobile computing. In J2MPP-Mine, they first

propose a subset finder strategy named Subset-Finder

(S-Finder) to find the possible subsets for prune.

Then, they propose a Subset pruner algorithm

(SBPruner) for determining the frequent pattern.

Furthermore, they proposed the novel prediction

strategy to determine the superset and remove the

subset which generates a less number of sets due to

different filtering pruning strategy. Finally, through

the simulation their proposed methods were shown to

deliver excellent performance in terms of efficiency,

accuracy and applicability under various system

conditions. Means if optimization is achieved the

data mining is also easily supported on mobile

devices.

In 2012, Nikhil Jain et al. [12] discuss about

Association rule mining. They suggest that

association rule play important rule in market data

analysis and also in medical diagnosis of correlated

problem. For the generation of association rule

mining various technique are used such as Apriori

algorithm, FP-growth and tree based algorithm. Some

algorithms are wonder performance but generate

negative association rule and also suffered from

Superiority measure problem. They proposed a multi-

objective association rule mining based on genetic

algorithm and Euclidean distance formula. In this

method we find the near distance of rule set using

Euclidean distance formula and generate two class

higher class and lower class .the validate of class

check by distance weight vector.

In 2012, Ashutosh Dubey et al. [20] Proposes an

efficient method for knowledge discovery which is

based on subset and superset approach. In this

approach they also use dynamic minimum support so

that we reduce the execution time. A frequent

superset means it contains more transactions then the

minimum support. It utilize the concept that if the

item set is not frequent but the superset may be

frequent which is consider for the further data mining

task. By this approach we can also find improved

association, which shows that which item set is most

acceptable association with others. A frequent subset

means it contains less transactions then the minimum

support. It utilizes the behavior that the less count

may be frequent if we attached the less count with the

higher order set. Here we also provide the flexibility

to find multiple minimum supports which is useful

for comparison with associated items and dynamic

support range. Their algorithm provides the

flexibility for improved association and dynamic

support. Comparative result shows the effectiveness

of their algorithm.

In 2012, Preeti Khare et al. [21] discusses that

importance of data mining is increasing exponentially

since last decade and in recent time where there is

very tough competition in the market where the

quality of information and information on time play a

very crucial role in decision making of policy has

attracted a great deal of attention in the information

industry and in society as a whole. They use density

minimum support so that they reduce the execution

time. A frequent superset means it contains more

transactions then the minimum support. It utilize the

concept that if the item set is not frequent but the

superset may be frequent which is consider for the

further data mining task. By this approach they can

store the transaction on the daily basis, then they

provide three different density zone based on the

transaction and minimum support which is low(L),

Medium(M),High(H). Based on this approach they

categorize the item set for pruning. Their approach is

based on apriori algorithm but provides better

reduction in time because of the prior separation in



114

the data, which is useful for selecting according to

the density wise distribution in India.

In 2012, Leena A Deshpande et al. [22] discusses

about Semi-structured data which are a huge amount

of complex and heterogeneous data sets. Such models

capture data that are not intentionally structured, but

are structured heterogeneously. These databases

evolve so quickly like run time report generated by

ERPs, World-Wide Web with its HTML pages, text

files, bibliographies, various logs generated etc.

These huge and varied become difficult to retrieve

relevant information User is often interested in

integrating various formats (like in biomedical data

text, image or structured) that are generally realized

as files, and also wants to access them in an

integrated fashion. Users not only query the data to

find a particular piece of information, but he is also

keen in knowing better understanding of the query.

Because of this variety, semi-structured DBs do not

come with a conceptual schema.

In 2012, Smruti Rekha Das et al. [23] discusses about

Support vector machine (SVM) which has become an

increasingly popular tool for machine learning tasks

involving classification, regression or novelty

detection. SVM is able to calculate the maximum

margin (separating hyper-plane) between data with

and without the outcome of interest if they are

linearly separable. To improve the generalization

performance of SVM classifier optimization

technique is used. According to the authors

Optimization refers to the selection of a best element

from some set of available alternatives. Particle

swarm optimization (PSO) is a population based

stochastic optimization technique where the potential

solutions, called particles, fly through the problem

space by following the current optimum particles.

They used Principal Component Analysis (PCA) for

reducing features of breast cancer, lung cancer and

heart disease data sets and an empirical comparison

of kernel selection using PSO for SVM is used to

achieve better performance. This paper focused on

SVM trained using linear, polynomial and radial

basis function (RBF) kernels and applying PSO to

each kernels for each data set to get better accuracy.

In 2012, Sanat Jain et al. [24] present an Apriori-

based algorithm that is able to find all valid positive

and negative association rules in a support confidence

framework. The algorithm can find all valid

association rules quickly and overcome some

limitations of the previous mining methods. The

complexity and large size of rules generated after

mining have motivated researchers and practitioners

to optimize the rule, for analysis purpose. Their

optimization done using Genetic Algorithm.

5. Analysis

After analyzing several research works in this

direction, we come with following analysis:

1) Optimize association rules are also useful in

the case of mobile data mining.

2) We can apply ACO and PSO for rule

optimization.

3) Genetic algorithm is already applied in case

of rule optimization [24].

4) Data partitioning is also needed, so that it

can maintain the partitioning speed.

5) Subset and superset based distribution is

also useful in the case of rule reduction.

6. Conclusion and Future Suggestions

In this paper we survey about association rule

mining, negative and positive rule generation and

optimization. We discuss different optimization

methods. We also survey related research in the

direction of rule optimization and provide the

analysis. Based on this we suggest some further

suggestions which are as follows:

1) Rule optimization can be applied using ACO

and PSO and compare with the genetic

algorithm.

2) After applying rule optimization, we can

obtain reduced rules which are helpful in the

case of handheld device.

3) Rule optimization is also done after

partitioning.

References

[1] J.Z.Mohammed.Parallel and distributed data mining: an

introduction. M.J.Zaki, C.-T.Ho (Eds.) Large-scale parallel

data mining, Lecture Notes in Artificial Intelligence,

1759:1-23, 2000.

[2] Sweeney, L. k-anonymity: a model for protecting

privacy. International Journal on Uncertainty, Fuzziness

and Knowledge-based Systems, 10(5):557-570, 2002.

[3] Agrawal, R. and Srikand R. Privacy preserving data

mining. In Proc. Of ACM SIGMOD Conference, pp. 439-

450, 2000.

[4] Chen, K. and Liu. L. A random rotation perturbation approach to privacy data classification. In Proc of IEEE

Intl. Conf. on Data Mining (ICDM), pp. 589-592, 2005.



115

[5] Xu, S., Zhang, J., Han, D. and Wang J. Singular value

decomposition based data distortion strategy for privacy

distortion. Knowledge and Information System, 10(3):383-

397, 2006.

[6] Mukherjeee, S., Chen, Z. and Gangopadhyay, A. A

privacy-preserving technique for Euclidean distance-based

mining algorithms using Fourier related transforms. Journal

of VLDB, 15(4):293-315, 2006.

[7] Vaidya, J. and Clifton, C. Privacy preserving k-means

clustering over vertically partitioned data. In Prof. of ACM

SIGKDD Conference, pp.206-215, 2003.

[8] Vaidya, J., Yu, H. and Jiang, X. Privacy preserving SVM classification. Knowledge and Information Systems,

14:161-178, 2007.

[9] Ms. Kumudbala Saxena, Dr. C.S. Satsangi, “A Non

Candidate Subset-Superset Dynamic Minimum Support Approach for sequential pattern Mining”, International

Journal of Advanced Computer Research (IJACR),Volume-

2 Number-4 Issue-6 December-2012.

[10] Dr. Manish Shrivastava, Mr. Kapil Sharma, MR. Angad Singh, “Web Log Mining using Improved Version

of Proposed Algorithm”, International Journal of Advanced

Computer Research (IJACR),Volume 1 Number 2

December 2011.

[11] Pragati Shrivastava, Hitesh Gupta,” A Review of

Density-Based clustering in Spatial Data”, International

Journal of Advanced Computer Research (IJACR),

Volume-2 Number-3 Issue-5 September-2012.

[12] Nikhil Jain,Vishal Sharma,Mahesh Malviya,

“Reduction of Negative and Positive Association Rule

Mining and Maintain Superiority of Rule Using Modified

Genetic Algorithm”, International Journal of Advanced

Computer Research (IJACR) ,Volume-2 Number-4 Issue-6 December-2012.

[13] S. Goss, S. Aron, J. L. Deneubourg, and J. M. Pasteels.

Self-organized Shorcuts in the Argentine Ant.

Naturwissenschaften, 76:579–581, 1989.

[14] M. Dorigo, Gianni Di Caro, and Luca M.

Gambardella. Ant Algorithms for Discrete Optimization.

Technical Report Tech. Rep. IRIDIA/98-10, IRIDIA,

Universite Libre de Bruxelles, Brussels, Belgium, 1998.

[15] M. Dorigo and M. Maniezzo and A. Colorni. The Ant

Systems: An Autocatalytic Optimizing Process. Revised

91-016, Dept. of Electronica, Milan Polytechnic, 1991.

[16] M. Dorigo and G. Di Caro. New Ideas in Optimisation.

McGraw Hill, London, UK, 1999.

[17]J. Kennedy, R. Eberhart . Particle swarm optimization.

In Proceedings of IEEE international conference on neural networks,Piscataway, NJ, USA (pp. 1942–1948),1995.

[18] Priyanka Gonnade, Sonali Bodkhe, “An Efficient load

balancing using Genetic algorithm in Hierarchical

structured distributed system”, International Journal of

Advanced Computer Research (IJACR),Volume-2

Number-4 Issue-6 December-2012.

[19] Ashutosh K. Dubey and Shishir K. Shandilya,” A

Novel J2ME Service for Mining Incremental Patterns in

Mobile Computing”,Communications in Computer and

Information Science, 2010,Springer LNCS.

[20] Ashutosh Kumar Dubey, Animesh Kumar Dubey,

Vipul Agarwal, Yogeshver Khandagre, “Knowledge

Discovery with a Subset-Superset Approach for Mining

Heterogeneous Data with Dynamic Support”,Conseg-2012.

[21] Preeti Khare, Hitesh Gupta, “Finding Frequent Pattern

with Transaction and Occurrences based on Density

Minimum Support Distribution”, International Journal of

Advanced Computer Research (IJACR),Volume-2 Number-3 Issue-5 September-2012.

[22] Leena A Deshpande, R.S. Prasad, “Efficient Frequent

Pattern Mining Techniques of Semi Structured data: a

Survey”, International Journal of Advanced Computer Research (IJACR) Volume-3 Number-1 Issue-8 March-

2013.

[23] Smruti Rekha Das, Pradeepta Kumar Panigrahi,

Kaberi Das and Debahuti Mishra, “Improving RBF Kernel Function of Support Vector Machine using Particle Swarm

Optimization”, International Journal of Advanced

Computer Research (IJACR) Volume-2 Number-4 Issue-7

December-2012.

[24] Sanat Jain, Swati Kabra, “Mining & Optimization of

Association Rules Using Effective Algorithm”, International Journal of Emerging Technology and

Advanced Engineering, Volume 2, Issue 4, April 2012.

association rules optimization: a survey - semantic · pdf fileassociation rules optimization:...

Documents